VIII' SAMPLING

1
VIII. SAMPLING
2
A. Overview

• 1. definition the process of selecting
  elements that
• are likely to reflect the
  variations that
• exist in the population
  of concern
• gt a method of identifying those units
  of analysis that
• reflect the diversity of any
  population without
• measuring the entire population
• 2. sometimes measuring an entire
  aggregate from any
• one population is unlikely or
  impossible
• 3. at issue is whether or not the
  elements selected are
• typical
• 4. the selection strategy must be
  stipulated, identified

3
B. Sampling Concerns

• 1. sampling bias
• a. are the respondents / responses you
  gather typical
• or unique?
• gt i.e., are your data reasonable?
• b. presence of social desirability
• gt hermeneutics and the Hawthorne
  Effect
• c. a question of data validity
• gt G I G O (garbage in, garbage out)

4

• 2. representativeness
• a. do those elements selected for
  inclusion into your
• research actually reflect those
  from whom they
• were selected?
• b. i.e., will data
  gathered from Criminal Justice
• majors about a topic of concern
  reflect that which
• might be gathered from any other
  campus major?
• c. if sample aggregated data
  approximates the
• population sample is probably
  representative

5

• 3. basic sampling principle
• a sample will be representative, IF
  all the
• elements from the population it was
  drawn have
• an equal opportunity of being included
  into the
• sample

6
C. Sampling Terminology

• 1. population the aggregate / total
  number of elements
• intended for study
• 2. element unit of analysis
• 3. parameter a summary description of a
  given
• variable from a
  population
• 4. statistic a summary description of a
  given variable
• from a sample
• 5. sampling error the discernable
  difference
• between a
  parameter and a statistic

7

• 6. random having no identifiable
  pattern or direction
• haphazard, irregular,
  arbitrary
• gt stochastic processing

8
D. Types of Probability Samples

• 1. overview
• a. the implementation of the basic
  sampling principle
• b. used whenever possible as it
  increases the validity
• of the statistical analysis techniques
  favored in
• social science analysis
• gt a basic assumption supporting
  use of general
• linear model statistics
• c. all strategies require a
  comprehensive list of
• elements

9

• 2. types of probabilty samples
• a. simple random sampling
• 1) the talk of most
  research projects, the reality
• of few
• 2) each element and stage is
  randomly selected
• b. systematic random sampling
• 1) this is what lay researchers
  think of as
• random sampling
• 2) selection of every Kth
  element from a
• population

10

• 4. stratified random sampling
• a. resembles a non-probability quota
  sample
• b. divide target population into
  identifiable / specific
• homogeneous subgroups /
  categories, and then
• select elements from each
  subgroup are randomly
• selected
• c. key is to insure that every
  subgroup is represented

11

• 5. multistage cluster sampling
• a. can be used even if comprehensive
  list of
• population is unavailable
• b. as with stratified sampling,
  elements are arranged
• into their natural clusters /
  subgroups found in the
• population, or as directed by the
  research
• project
• c. then entire clusters are randomly
  selected

12
E. Types of Non-Probability Samples

• 1. overview
• a. when randomization is not possible,
  impossible or even
• undesirable
• b. when it is better to have specific
  elements from a population
• rather than any
• 2. types on non-probability samples
• a. quota sampling
• 1) based on proportionality of
  the population of concern
• 2) selection of elements, not
  randomly, to satisfy preconceived
• categories / elements of the
  target population

13

• b. snowball sampling
• 1) based on availability of
  elements
• 2) use of informants
• 3) ethnographic research
• c. deviant case sampling
• 1) a specific form of snowball
  sampling
• 2) when only specific elements
  from a population
• are required
• d. convenience / judgmental /
  purposivesampling
• 1) the preferred form of the
  amateur

14
F. Summary

• 1. the raison detre of sampling is to obtain
  valid sources of
• information when measuring entire
  populations is impossible /
• unlikely
• 2. the key is to obtain representative
  elements
• 3. remember, these will be the exact sources
  of the theory you
• are testing
• 4. probability samples, especially SRS are
  the preferred
• strategies, often not the realistic ones
• 5. non-probability samples are less reliable
  than probability
• samples but provided greater validity
• gt are often more representative

15

• 6. cautions
• a. randomization
• gt because of the statistical
  techniques available,
• need is to sample as close to
  randomly as possible
• b. using college student samples as
  representatives of
• the general population sets up
  artificial analysis
• c. when conducting true / classic
  experiments, matched
• pair samples is imperative
• 7. in the end you get what you get and be
  happy with it